Nvidia RTX Spark Superchip: Specs, Price, and Release Date

Nvidia just entered the personal computer market. At Computex 2026 on June 1, Jensen Huang unveiled the RTX Spark — an Arm-based superchip designed to compete head-on with Apple's M5. It combines a 20-core Grace CPU with Nvidia's Blackwell GPU, up to 128GB of unified memory, and delivers 1 petaflop of on-device AI performance. Devices from ASUS, Dell, HP, Microsoft, Lenovo, and MSI ship this fall, with Acer and Gigabyte to follow.

Key Takeaways

• RTX Spark pairs a 20-core Arm CPU (co-designed with MediaTek) and a Blackwell GPU with 6,144 CUDA cores
• Up to 128GB LPDDR5X unified memory on TSMC's 3nm node — same approach Apple uses with M-series
• Benchmarks show it beats the base M5 by 54% in CPU throughput, but trails the M5 Pro
• Battery life: 17 hours of 4K video playback — two hours more than the M4 Max MacBook Pro
• Devices expected around $2,000+, shipping Fall 2026 (September–November window)

What the RTX Spark Superchip Actually Is

The RTX Spark (internally also called the N1X) is Nvidia's first processor designed for consumer Windows PCs. This isn't a discrete GPU bolted into a laptop — it's a fully integrated system-on-chip built from the ground up.

The core specs:

• CPU: 20-core Grace Arm design, co-engineered with MediaTek
• GPU: Blackwell architecture, 6,144 CUDA cores, fifth-generation Tensor cores
• Memory: Up to 128GB LPDDR5X coherent unified memory
• Interconnect bandwidth: 600 GB/s via Nvidia NVLink-C2C
• Process node: TSMC 3nm
• AI performance: 1 petaflop FP4, fully on-device

Nvidia says the integrated GPU is roughly equivalent to a desktop RTX 5070. That's a large claim for a laptop chip with a thin thermal envelope. The chip also supports DLSS 4.5, meaning more than 1,000 RTX-compatible games and apps run natively — the largest launch compatibility library of any new PC platform.

For developers and AI researchers, the 128GB unified memory pool is the more important headline. Running 70B+ parameter language models locally without cloud infrastructure becomes realistic. This is the same strategy Apple demonstrated with M-series silicon, just arriving later on Windows.

Tom's Hardware has the most detailed technical breakdown of the chip architecture for those who want to go deeper.

How RTX Spark Benchmarks Against Apple Silicon

Early Clang compiler benchmarks (multi-threaded CPU throughput) show RTX Spark scoring 43,149 versus the base M5's 27,996 — a 54% lead over Apple's entry-level chip. That's the most flattering comparison for Nvidia, and the numbers shift once you move up Apple's lineup.

The M5 Pro (15-core) scores 46,374, putting RTX Spark about 7% behind. The 18-core M5 Pro widens that gap to roughly 22%. RTX Spark doesn't aim to beat Apple's entire lineup on CPU throughput — it aims to match the mainstream MacBook Pro tier while adding GPU capability Apple simply can't match.

Chip	CPU Cores	Clang Score	Delta vs RTX Spark
Nvidia RTX Spark	20 (Arm/Grace)	43,149	—
Apple M5 (base)	10	27,996	−35%
Apple M5 Pro 15c	15	46,374	+7%
Apple M5 Pro 18c	18	55,165	+28%

Where RTX Spark clearly wins: gaming and GPU-accelerated workloads. Apple's integrated GPU can't compete with 6,144 Blackwell CUDA cores for real-time 3D rendering, video encoding with NVENC, and FP4 AI inference. The chip is targeting creators and power users who need local AI and serious GPU throughput in one thin laptop — a different buyer profile than the M5 base model, according to TechRadar's Computex coverage.

Which Laptops Are Coming (and When to Expect Them)

Nvidia confirmed eight OEM partners for the RTX Spark launch wave, all targeting a Fall 2026 window (September–November):

• ASUS — ProArt P14 and P16, plus a compact mini PC
• Microsoft — Surface Laptop Ultra (their first Arm Windows flagship with Nvidia silicon)
• HP — OmniBook lineup
• Dell — XPS series response to MacBook Neo
• Lenovo — ThinkPad and IdeaPad variants (SKUs TBA)
• MSI — Creator and gaming configurations
• Acer — Models following shortly after initial launch
• GIGABYTE — Following after initial wave

Nvidia expects to ship more than 30 laptop models and 10 desktop configurations across all partners. The desktop form factor targets the mini PC category — compact, fanless or near-silent machines for developers running local AI inference workloads.

Pricing isn't confirmed for consumer models, but the professional DGX Spark (same underlying chip) starts at $3,299 for the 64GB config. Consumer RTX Spark laptops are expected to start closer to $2,000 given volume manufacturing, though premium configs with 128GB will likely climb significantly.

Nvidia also revealed a three-generation roadmap: RTX Spark (Grace + Blackwell) is the first generation, followed by Vera Rubin Spark with LPDDR6 memory, then Rosa Feynman Spark with next-generation stacked GPU architecture.

Why Nvidia Is Entering the PC Market Now

The timing isn't accidental. AI workloads are moving from data centers to personal devices, and Windows has lacked a competitive unified-memory chip for this transition. Apple's M-series proved the market exists — Nvidia is betting it can capture the Windows side of it.

Jensen Huang described RTX Spark as reinventing "the most important tool in the history of humanity." The market agreed quickly: NVDA stock jumped over 6% on the announcement.

For developers planning ahead for RTX Spark hardware, check out the temp mail for developers guide — when you're registering for OEM beta programs, Nvidia developer access, or early-adopter waitlists across multiple partner sites before devices ship, a disposable inbox from app.fasttempmail.com keeps your main address clean from the marketing follow-up that inevitably follows.

The broader PC market shift is real. If RTX Spark delivers on its benchmarks in retail hardware — particularly battery life and AI inference speed — it's the most credible challenge to Apple Silicon that Windows has produced. The real test comes when devices ship this fall.

FAQ

What is the Nvidia RTX Spark superchip?

The Nvidia RTX Spark is an Arm-based system-on-chip announced at Computex 2026. It integrates a 20-core Grace CPU, a Blackwell GPU with 6,144 CUDA cores, and up to 128GB of unified LPDDR5X memory, designed for AI workloads, gaming, and content creation on Windows PCs.

How does Nvidia RTX Spark compare to Apple M5?

RTX Spark beats the base M5 (10-core) by 54% in CPU throughput benchmarks, but trails the M5 Pro in raw CPU performance. In GPU-intensive tasks and AI inference, RTX Spark's Blackwell architecture and 1 petaflop FP4 performance exceed what Apple's integrated GPU can handle.

When will RTX Spark laptops be available?

RTX Spark devices are expected to ship Fall 2026 — between September and November — from partners including ASUS, Dell, HP, Lenovo, Microsoft, and MSI. Prices are expected to start above $2,000 for entry configurations.

Does Nvidia RTX Spark work with Windows games?

RTX Spark supports over 1,000 RTX-compatible games and apps natively, with DLSS 4.5 included at launch. The integrated Blackwell GPU with 6,144 CUDA cores is roughly equivalent to a desktop RTX 5070, targeting 100+ fps at 1440p in AAA titles.